# Backpropagation **Domain:** Machine Learning / Optimization / Neural Networks **Doc Type:** Technical Concept Node **Maturity:** Developed **Related:** [[wiki/Artificial Neural Networks|Artificial Neural Networks]], [[wiki/Geoffrey Hinton|Geoffrey Hinton]], [[wiki/Yann LeCun|Yann LeCun]] --- ## Definition **Backpropagation** is the efficient application of the chain rule for calculating how a network's error changes with respect to its parameters. Those gradients allow an optimization procedure to adjust weights throughout a multilayer computational graph. ## Ontology Context Backpropagation is a training calculation, not an intelligence theory and not a complete learning system. It routes to [[wiki/Artificial Neural Networks|Artificial Neural Networks]] for the trained architecture and to the relevant person nodes for historical research lineages. ## See Also [[wiki/Artificial Neural Networks|Artificial Neural Networks]] · [[wiki/Geoffrey Hinton|Geoffrey Hinton]] · [[wiki/Yann LeCun|Yann LeCun]] · [[wiki/Deep Learning|Deep Learning]]